Bug 249038 - [WebAssembly SIMD] Emulate vector floating-point absolute value on Intel
Summary: [WebAssembly SIMD] Emulate vector floating-point absolute value on Intel
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebAssembly (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: David Degazio
URL:
Keywords: InRadar
Depends on:
Blocks: 246345
  Show dependency tree / graph
 
Reported: 2022-12-09 11:55 PST by David Degazio
Modified: 2022-12-13 11:48 PST (History)
0 users

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description David Degazio 2022-12-09 11:55:12 PST 
Intel doesn't have an instruction for SIMD floating-point absolute value, so the WASM f32x4.abs and f64x2.abs opcodes are currently unsupported. We can emulate this feature by generating a vector bitmask and using it to clear the sign bits for all elements in a vector.
Comment 1 David Degazio 2022-12-09 11:56:52 PST 
rdar://103189245
Comment 2 David Degazio 2022-12-09 12:40:46 PST 
Pull request: https://github.com/WebKit/WebKit/pull/7402
Comment 3 EWS 2022-12-13 11:48:24 PST 
Committed 257806@main (e2d9691a6968): <https://commits.webkit.org/257806@main>

Reviewed commits have been landed. Closing PR #7402 and removing active labels.